1. Field of the Invention
Embodiments of the present invention generally relate to 3D modeling and, more specifically, to generating informative viewpoints based on editing history.
2. Description of the Related Art
The presentation of a three-dimensional (3D) model in a venue such as an online forum or the sharing of a 3D model with a manager or a peer for review is common practice for an author of a 3D model. However, sharing the 3D model in the same format used to create or edit the 3D model is not always convenient. The file containing the 3D model may be too large to transmit to a reviewer easily, the reviewer may not have access to the correct software program for opening and viewing the 3D model, or the reviewer might find navigating the 3D model difficult and time consuming. There may also be piracy concerns involved with presenting the 3D model in any type of open or non-secure forum. Therefore, the author of the 3D model oftentimes generates and distributes collages or collections of two-dimensional (2D) renderings of the 3D model to the reviewers. Typically, the 2D renderings or collages are easier to transmit, may be viewed with standard desktop computer software, and limit the author's exposure to piracy.
In order to represent the 3D model via a collection of 2D renderings with reasonable accuracy, the 2D renderings should be created from previously determined viewpoints of the 3D model that show important features of the 3D model. To select viewpoints from which informative representations of those features may be captured (i.e. “informative” viewpoints), the author usually has to explore each such feature from many different viewpoints. Rotating, panning, and zooming around the 3D model to identify and then select these informative viewpoints requires complex user input, which is oftentimes tedious and time-consuming. This navigation of the 3D model oftentimes makes the manual creation of a collage of 2D renderings an overly burdensome process. Therefore, systems to automate the selection of viewpoints of 3D models have been developed to assist in the creation of collages of 2D renderings.
Existing systems for automating the selection of informative viewpoints attempt to identify important features of the 3D model and then select viewpoints of the 3D model that show the identified features. Generally speaking, these systems correlate the importance of a feature with either the amount of time a reviewer of the completed 3D model spent viewing the feature or with the geometric complexity of the feature.
One drawback of selecting viewpoints based upon the behavior of a reviewer is that a reviewer must manually explore a completed 3D model, which is time-consuming and difficult for a reviewer. The amount of time required for a reviewer to navigate the 3D model is comparable to the time required to select informative viewpoints manually, so systems based upon the behavior of a reviewer provide little advantage over creating a collage manually.
One drawback of selecting viewpoints based upon geometric complexity is that complexity does not necessarily equate to importance. For example, techniques exist in the field of 3D modeling that are capable of producing geometrically complex regions that may be of little interest to a viewer of the 3D model. In applying such a technique, an author of a 3D model of a person may sculpt hair by quickly scribbling across the surface of the head of the 3D model. The hair would have a high-level of geometric complexity but would not be of much importance in the overall presentation of the person. A system configured to select viewpoints based on geometric complexity would likely select viewpoints of different patches of hair, but not of the face or body of the person. Consequently, a collage of 2D renderings of only the hair of the person would not properly represent the entirety of the modeled person and would not capture the most important features of the modeled person.
As the foregoing illustrates, what is needed in the art is a more effective way to generate informative viewpoints for a 3D model when creating a collage of 2D renderings that represent the 3D model.